Physics engine on reconfigurable processor — Low power optimized solution empowering next-generation graphics on embedded platforms

Physics engine is an important auxiliary middleware component library needed by the graphics display system. It is used to provide realistic animation effects in applications ranging from games, multimedia, user interfaces etc based on Newtonian laws of physics. With the evolution of high-end graphics on embedded platforms, there is a subtle need for the provision of a dedicated physics library whose processing can be offloaded to separate processing unit other than the main processor to maintain the optimal frame-rate requirement of the end user application. Currently physics processing is accelerated using high power high performance architectures like graphics processing units (GPU) or physics processing units (PPU). Reconfigurable computing facilitates execution of software with higher flexibility than that of any dedicated hardware while providing increased performance and low power consumption. In this paper we demonstrate that physics processing can be optimized using a reconfigurable processor. We also propose a novel method of executing collision solver algorithm in a physics engine library on Coarse Grained Reconfigurable Array (CGRA) based reconfigurable processors.